LAUSR

Abstract Controversies surround the origin of kimberlite megacrysts, including whether and how they are genetically related to their host kimberlites, the relationship between the Cr-poor and Cr-rich suites and the dominant processes responsible for elemental and isotopic variations of megacrysts from a given kimberlite. We present new in-situ major and trace element and Sr isotopic results for clinoyroxene and garnet megacrysts from four southern African kimberlites: Colossus and Orapa (Group 1 kimberlites on the Zimbabwe craton), and Kalkput and Bellsbank (Group 2 kimberlites on the western Kaapvaal craton), that include both Cr-poor and Cr-rich megacryst varieties. Cr-poor megacrysts are present at Colossus, Orapa and Kalkput and the data exhibit tight, well-defined trends on major element diagrams as well as incompatible and rare earth element abundances similar to those previously reported for Cr-poor megacrysts. Cr-rich megacrysts, which are also present at Orapa and are the only variety present at Bellsbank, generally have higher Mg# values, lack well-defined major element trends and show stronger incompatible element enrichments as well as more radiogenic 87Sr/86Sri ratios than Cr-poor megacrysts from the same kimberlite group. Thermobarometry indicates that the Cr-poor megacrysts equilibrated at temperatures of ≈1200 to 1450 °C and pressures of 4.5 to 7.5 GPa. Cr-rich megacrysts, in contrast, extend to temperatures and pressures as low as 700 °C and 3 GPa, respectively. This indicates that, in the studied suites, Cr-poor megacrysts equilibrated at high temperatures in the lower lithosphere (>135 km), whereas Cr-rich megacrysts typically equilibrated at lower temperatures and pressures. Within the Cr-poor megacrysts from Group 1 and Group 2 kimberlites, there is a clear correspondence between kimberlite group, diagnostic incompatible element ratios (e.g., Nb/La) and Sr isotope ratios that parallel the differences noted between whole-rock Group 1 and Group 2 kimberlites. In the case of Cr-poor megacrysts, similar calculated melt compositions in equilibrium with garnet and clinopyroxene from the same kimberlite were obtained using recent high-pressure mineral‑carbonated melt partition coefficients. This suggests formation in conditions close to trace element equilibrium, and is consistent with crystallization from primitive melts with kimberlite-like trace element compositions. In the case of Cr-rich megacrysts, differences in the compositions of melts in equilibrium with clinopyroxene and garnet tend to be larger, and melts in equilibrium with Cr-rich clinopyroxene tend to show significantly greater incompatible element enrichments than those of estimated near-primary kimberlite melts. This could be due to the different behaviour of clinopyroxene and garnet during metasomatic melt-rock interaction, but the apparent disequilibrium between clinopyroxene and garnet could also be due to some of the Cr-rich megacrysts actually being peridotitic xenocrysts. We propose a model for the origin of southern African megacrysts in which carbonated protokimberlite melts formed stockwork-like bodies of variable size in the deep lithosphere (>130 km), which fed networks of melt-filled veins extending into the surrounding and overlying mantle. Crystallization of larger melt bodies resulted in megacryst assemblages dominated by Cr-poor megacrysts, and the incompatible element and isotopic characteristics of these dominantly reflect those of the protokimberlite melt. In contrast, crystallization of smaller melt bodies and their vein networks resulted in megacryst assemblages dominated by Cr-rich megacrysts, which formed as a result of extensive assimilation and metasomatic melt-rock interaction between protokimberlite and peridotite wallrock at low melt/rock ratios, particularly in the middle to shallow lithosphere where pre-existing potassic metasomatic heterogeneities are prevalent. The Cr-rich nature and enrichments in incompatible elements and radiogenic Sr in the Cr-rich megacrysts reflect extensive interaction of their parental magmas with this metasomatized peridotite.